Control system of medical multifunctional mobile slide pad

Abstract

The application belongs to the field of electric digital data processing, and discloses a control system of a medical multifunctional mobile sliding pad. The system obtains dynamic pressure distribution data through a matrix type flexible pressure sensor network and organizes the data into a multi-dimensional pressure matrix data structure. An edge computing module filters and interpolates the data, extracts the pressure area and pressure duration through eight-connected domain marking, calculates the pressure centroid to construct a velocity gradient matrix and extract the sliding displacement trend. According to the modeling analysis result, the system generates a three-dimensional characteristic tensor and maps it into a switch state sequence and a charging and discharging duration sequence of a micro air bag electromagnetic valve array, executes a partition charging or discharging operation, and issues an alarm data packet in an abnormal situation. The system solves the problem that local pressure area expansion or slight sliding cannot be identified, and realizes automatic balanced adjustment of the stress surface.

Description

Technical Field

[0001] This invention belongs to the field of electronic digital data processing and discloses a control system for a medical multifunctional mobile sliding pad. Background Technology

[0002] Medical mobile sliding mats are typically used as auxiliary tools for patient relocation. Existing sliding mats with monitoring functions usually have discrete piezoresistive sensors embedded inside. In conventional solutions, a microprocessor scans the resistance values ​​of these discrete sensors one by one at a fixed frequency. When the resistance value of a single sensor exceeds a preset fixed threshold, the microprocessor sends a single byte of status change level to the nursing host via a wireless transmission module. After receiving the status change level, the nursing staff manually adjusts the external air pump to inflate or deflate the corresponding area of ​​the sliding mat according to the sensor number, or manually resets the patient who has slipped. At the data interaction level, conventional solutions simply concatenate the sensor's physical number and trigger status and send it to the receiving end.

[0003] The conventional technical solutions described above rely on a fixed threshold triggering mechanism of a single discrete sensor, which cannot obtain the overall force distribution on the patient's contact surface. When the patient experiences minor slippage or the local pressure area gradually expands, the pressure changes caused by the slippage or area expansion are distributed across multiple adjacent discrete sensors. This results in the pressure change at the location of a single sensor not reaching the preset fixed threshold. Consequently, the system cannot identify the risk of high-pressure sores or abnormal slippage caused by minor slippage or local area expansion, and therefore cannot output the corresponding airbag adjustment command in a timely manner to automatically balance and adjust the force-bearing surface. Summary of the Invention

[0004] The purpose of this invention is to provide a control system for a medical multifunctional mobile sliding pad, which can effectively solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: The control system for the medical multifunctional mobile sliding pad includes a high-resolution matrix flexible pressure sensor network, a microcontroller, an edge computing module, a wireless gateway, and a micro airbag solenoid valve array laid inside the sliding pad. The microcontroller continuously collects dynamic pressure distribution data of the patient contact surface output by the high-resolution matrix flexible pressure sensor network using a preset time polling mechanism. The edge computing module has a built-in posture recognition algorithm, receives the dynamic pressure distribution data, and performs data modeling and in-depth analysis on the patient's pressure area, pressure duration, and slippage trend. When a risk of pressure sores is detected or an unexpected abnormal slip occurs, the edge computing module sends an alarm data packet to the nursing station through the wireless gateway, and at the same time outputs a feedback control signal based on the force model. The control system of the medical multifunctional mobile sliding pad, which receives the feedback control signal from the micro airbag solenoid valve array, performs zoned inflation or deflation operations.

[0006] Preferably, the high-resolution matrix flexible pressure sensor network is composed of flexible pressure sensing nodes with alternating rows and columns, and each flexible pressure sensing node is assigned a unique row and column address identifier. The dynamic pressure distribution data is organized into a multi-dimensional pressure matrix data structure, which includes a frame header, timestamp, row number identifier, column number identifier, and a pressure value sequence arranged in row priority order. The frame header includes a data length field and a check field. The timestamp is the absolute time when the microcontroller reads the flexible pressure sensor node. Each element in the pressure value sequence corresponds to the unique row and column address identifier.

[0007] Preferably, the microcontroller includes a polling schedule table and a dynamic frequency register. The polling schedule table divides the high-resolution matrix flexible pressure sensor network into multiple scanning sub-regions, and the dynamic frequency register stores the polling period value corresponding to each scanning sub-region. The time polling mechanism triggers reading operations on each scanned sub-region sequentially according to the polling schedule table, and performs data parsing tasks for other scanned sub-regions within the reading interval of a single scanned sub-region. When the rate of pressure change in any scanned sub-region exceeds a preset change threshold, the microcontroller updates the polling cycle value in the dynamic frequency register to shorten the polling cycle of the corresponding scanned sub-region.

[0008] Preferably, the posture recognition algorithm's data modeling process for the pressure area and pressure duration includes: binarizing the dynamic pressure distribution data to generate a pressure mask matrix, and performing an eight-connected domain labeling algorithm on the pressure mask matrix to extract at least one connected pressure region; For each connected compressed region, the compressed area is obtained by multiplying the number of internal pixels by the area of ​​a single pixel. Establish a duration accumulator corresponding to each of the connected pressure areas. Compare the overlap rate of the connected pressure areas in the dynamic pressure distribution data of multiple consecutive frames. When the overlap rate is greater than a set ratio, trigger the duration accumulator to count and accumulate to generate the pressure duration.

[0009] Preferably, the process of the attitude recognition algorithm performing in-depth analysis of the sliding displacement trend includes: extracting multi-frame dynamic pressure distribution data within a continuous time window, and calculating the global pressure centroid coordinates of each frame of dynamic pressure distribution data; The global pressure centroid coordinates of two adjacent frames are subtracted to generate a centroid displacement vector. Multiple consecutive centroid displacement vectors are arranged in chronological order to construct a centroid trajectory matrix. A velocity gradient matrix is ​​generated by performing a first-order difference operation on the centroid trajectory matrix, and continuous vector segments exceeding a preset velocity threshold in the velocity gradient matrix are extracted as slip displacement trend outputs.

[0010] Preferably, the alarm data packets generated by the edge computing module are assembled according to a preset encapsulation protocol, which defines a start symbol, a device identifier, an alarm type field, a feature load field, and an end symbol. The alarm type field is filled with a high pressure sore risk indicator or an abnormal sliding indicator, and the characteristic load field is filled with the corresponding value of the pressure area, the pressure duration, or the sliding displacement trend vector when the alarm is triggered. After receiving the alarm data packet, the wireless gateway matches the alarm type field with a preset transmission channel queue, and divides the alarm data packet into multiple fragmented data packets with sequence numbers for transmission.

[0011] Preferably, spatial filtering and spline interpolation reconstruction are performed on the pressure value sequence in the multidimensional pressure matrix data structure; The spatial filtering process constructs a sliding window of a preset size centered on the target element in the current pressure value sequence, calculates the spatial distance weight of each element within the sliding window, and uses the spatial distance weight to perform a weighted summation of the elements within the sliding window to generate filtered element values, which then replace the target element values. For missing positions in the filtered pressure value sequence caused by sensor node damage, cubic spline interpolation is performed using multiple valid element values ​​adjacent to the missing position as interpolation nodes, and the interpolation results are filled into the missing positions.

[0012] Preferably, the historical time-series pressure change rate is incorporated into the calculation during the process of the microcontroller updating the polling cycle value in the dynamic frequency register; The microcontroller maintains a circular queue of change rates corresponding to each scanning sub-region, pushes the pressure change rate obtained in the current polling cycle into the tail of the corresponding circular queue of change rates and removes the data at the head of the queue. A trend rate of change is generated by performing a moving average calculation on all pressure change rates in the rate of change cycle queue. The absolute value of the difference between the current pressure change rate and the trend rate of change is used to look up the frequency adjustment step size in the table. The polling period value in the dynamic frequency register is then modified by superimposing the frequency adjustment step size.

[0013] Preferably, a weight attenuation correction operation based on the geometric deformation parameters of the connected compressed region is added during the process of counting and accumulating in the duration accumulator to generate the compressed duration. For each connected pressure region, calculate the aspect ratio of the bounding rectangle of the corresponding region in the pressure mask matrix and the region area dispersion index; Construct a weighted decay function with aspect ratio and regional area dispersion index as independent variables, and multiply the calculation result of the weighted decay function as the decay coefficient into the single count value of the duration accumulator. When the aspect ratio deviates from the preset standard range or the area dispersion index increases, the single count value is reduced to correct the pressure duration.

[0014] Preferably, the sliding displacement trend and the characteristics of the pressure area are fused using tensors and mapped into control commands for the micro airbag solenoid valve array; The mean direction and magnitude of the continuous vector segments in the velocity gradient matrix are extracted, and combined with the geometric center coordinates of the pressure region extracted from the pressure mask matrix, a three-dimensional feature tensor containing directional features, magnitude features and position features is constructed. The three-dimensional feature tensor is input into a preset fully connected mapping network, and the output is a sequence of switching states and a sequence of inflation and deflation durations corresponding to each solenoid valve node in the micro airbag solenoid valve array, which serve as feedback control signals.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention acquires dynamic pressure distribution data of the patient's contact surface through a high-resolution matrix flexible pressure sensor network. A multi-dimensional pressure matrix data structure is used to organize the pressure value sequence with row and column address identifiers. An attitude recognition algorithm binarizes the dynamic pressure distribution data to generate a pressure mask matrix. An eight-connected domain labeling algorithm extracts connected pressure regions and statistically analyzes the pressure area and duration. Simultaneously, a centroid trajectory matrix is ​​constructed by calculating the global pressure centroid coordinates, and a velocity gradient matrix is ​​generated by performing a first-order difference operation to extract the slip displacement trend. Combining the modeling and analysis results of the pressure area, pressure duration, and slip displacement trend, the system can identify the risk of high-pressure sores or abnormal slip states caused by minute slippage or localized expansion of the pressure area. Furthermore, a three-dimensional feature tensor containing directional, amplitude, and positional features is mapped to the on / off state sequence and inflation / deflation duration sequence of a miniature airbag solenoid valve array, achieving automatic equalization adjustment of the force-bearing surface.

[0016] 2. During the data acquisition phase, scanning sub-regions are divided using a polling schedule. The polling cycle of each scanning sub-region is dynamically adjusted by combining a dynamic frequency register with a historical time-series pressure change rate cyclic queue, increasing the data acquisition density during sudden local pressure changes. During the data processing phase, a sliding window with spatial distance weights is used to perform spatial filtering on the pressure value sequence, and cubic spline interpolation is used to fill in missing positions, ensuring the integrity of the input data. For the cumulative calculation of pressure duration, a weighted attenuation function based on the aspect ratio of the circumscribed rectangle and the area dispersion exponent is introduced for correction, eliminating geometric deformation interference caused by patient turning. During the data transmission phase, alarm data packets are assembled according to a preset encapsulation protocol. The transmission channel queue is matched according to the alarm type, and the packets are segmented into fragmented data packets with sequence numbers for transmission, ensuring the reliability of abnormal state data transmission. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the overall operation of the medical multifunctional mobile sliding mat control system of the present invention. Figure 2 This is a flowchart of the dynamic pressure distribution data encapsulation and preprocessing process of the present invention; Figure 3 This is a flowchart illustrating the dynamic adjustment of the scanning sub-region polling cycle according to the present invention. Figure 4 This is a flowchart illustrating the calculation and correction process for the pressure area and pressure duration of the present invention. Figure 5 This is a flowchart of the slip displacement trend recognition and alarm data packet transmission of the present invention; Figure 6 This is a flowchart illustrating the generation of feedback control commands for the miniature airbag solenoid valve array of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please refer to Figure 1This embodiment provides a control system for a medical multifunctional mobile sliding pad, including a high-resolution matrix flexible pressure sensor network, a microcontroller, an edge computing module, a wireless gateway, and a miniature airbag solenoid valve array embedded within the sliding pad. The high-resolution matrix flexible pressure sensor network is laid between the patient contact surface layer and the buffer layer of the sliding pad, with the sensing surfaces of the sensing nodes facing the patient contact side, covering the entire effective contact area of ​​the sliding pad. It is used to sense pressure changes between the patient and the sliding pad contact surface and output corresponding electrical signals. The microcontroller is electrically connected to the high-resolution matrix flexible pressure sensor network via a row and column scanning bus. It continuously collects dynamic pressure distribution data of the patient contact surface output by the high-resolution matrix flexible pressure sensor network using a preset time polling mechanism. During the acquisition process, it performs analog-to-digital signal conversion and data verification to ensure the validity of the output data. The edge computing module establishes a data interaction channel with the microcontroller via a serial data bus. It has a built-in posture recognition algorithm, receives dynamic pressure distribution data uploaded by the microcontroller, and performs data modeling and in-depth analysis of the patient's pressure area, pressure duration, and sliding displacement trend. The edge computing module incorporates pre-defined risk assessment rules. Based on modeling and analysis results, when a risk of high pressure sores is detected or unexpected abnormal slippage occurs, the edge computing module transmits the generated alarm data packet to the wireless gateway via the local bus. The wireless gateway then distributes the alarm data packet to the receiving terminal at the nursing station. Simultaneously, the edge computing module outputs a feedback control signal based on a pre-defined force model. This feedback control signal is transmitted to the miniature airbag solenoid valve array via the control bus. Each solenoid valve node in the miniature airbag solenoid valve array corresponds to an independent airbag section within the sliding pad. After receiving the feedback control signal, the solenoid valve node performs inflation or deflation operations on the corresponding airbag section. By adjusting the inflation volume of the airbag section, the support height of the corresponding area is changed, thereby redistributing the pressure on the patient's contact surface.

[0020] Specifically, the sensing nodes of the high-resolution matrix flexible pressure sensor network employ piezoresistive sensing units. The resistance of these units changes linearly with the pressure applied to the sensing surface. The microcontroller converts these resistance changes into corresponding pressure values. The microcontroller's time-polling mechanism sequentially enables the row selection circuits of the sensor network according to a preset scanning order, synchronously reading the electrical signals of all column sensing units within the corresponding row to complete the acquisition of a single frame of dynamic pressure distribution data. The edge computing module's built-in attitude recognition algorithm preprocesses the received dynamic pressure distribution data, extracts the pressure distribution features of the patient's contact area, and quantifies the pressure area and duration through feature matching and statistical calculations. Simultaneously, it extracts the slip displacement trend through feature comparison of consecutive frame data. The risk assessment rule for high-pressure sores is as follows: when the pressure area of ​​a single connected pressure zone exceeds a preset area threshold, and the continuous pressure duration of that zone exceeds a preset duration threshold, a high-pressure sore risk is identified. The rule for unexpected abnormal slippage is as follows: when the magnitude of the displacement of the pressure centroid corresponding to consecutive frame data exceeds a preset displacement threshold, and the duration of the displacement trend exceeds a preset time threshold, unexpected abnormal slippage is identified. The feedback control signal generated by the edge computing module matches the location corresponding to the pressure zone triggering the risk or the slippage trend. For areas of concentrated pressure, an air release control signal is output to reduce the support height of the corresponding area to disperse pressure. For the displacement direction corresponding to the slippage trend, an inflation control signal is output for the corresponding area, limiting further patient slippage through the height difference. The wireless gateway and nursing station establish a communication connection via the hospital's wireless LAN. The transmission of alarm data packets follows the communication protocol specifications of the hospital's medical equipment to ensure the compliance and security of data transmission. Each solenoid valve node in the miniature airbag solenoid valve array is a two-position three-way solenoid valve, corresponding to the inflation path, deflation path, and pressure holding state, respectively. The solenoid valve node switches the path state according to the feedback control signal to realize the inflation, deflation, or pressure holding operation of the corresponding airbag zone.

[0021] In this embodiment, a corresponding mapping relationship is established between the partitions of the micro airbag solenoid valve array and the pressure sensing area. The mapping relationship is defined and stored through the following table.

[0022] Table 1 Mapping of Miniature Airbag Solenoid Valve Array Zones and Pressure Sensing Areas

[0023] The table above is stored in the storage unit of the edge computing module. The row and column address range in the table corresponds to the node row and column address identifiers of the high-resolution matrix flexible pressure sensor network. It is used to establish a precise matching relationship between the pressure abnormal area and the airbag control zone. When the edge computing module generates feedback control signal, it determines the solenoid valve node number that needs to perform inflation and deflation operations by querying this table, ensuring that the control signal can accurately act on the area corresponding to the pressure abnormality.

[0024] This embodiment constructs a complete operating architecture for a medical multifunctional mobile sliding mat control system. It acquires global dynamic pressure distribution data of the patient's contact surface through a matrix pressure sensor network, and combines it with a posture recognition algorithm to identify the risk of high pressure sores and abnormal sliding states. Simultaneously, it realizes the uploading of alarm information and the automatic adjustment of airbag zoning, solving the problems of recognition lag and inability to automatically adjust caused by relying on discrete sensors and fixed threshold triggering in the prior art.

[0025] In a preferred embodiment, reference Figure 2 The high-resolution matrix flexible pressure sensor network consists of rows and columns of flexible pressure sensing nodes, each assigned a unique row and column address identifier. All flexible pressure sensing nodes in a row share the same row gate line, and all flexible pressure sensing nodes in a column share the same column readout line. When the microcontroller enables a row gate line, all flexible pressure sensing nodes in that row output their current pressure sensing signal to the corresponding column readout line. The microcontroller completes the pressure data acquisition for all sensing nodes in that row by synchronously reading the signals from all column readout lines. The unique row and column address identifier for each flexible pressure sensing node is composed of a row number and a column number, both of which are consecutive integers starting from 1. The row number corresponds to the row gate line number where the sensing node is located, and the column number corresponds to the column readout line number where the sensing node is located, ensuring that the row and column address identifier of each sensing node is unique throughout the entire sensor network.

[0026] The dynamic pressure distribution data acquired by the microcontroller is organized into a multidimensional pressure matrix data structure. This structure includes a frame header, timestamp, row number identifier, column number identifier, and a pressure value sequence arranged in row-major order. The frame header contains a data length field and a check field. The data length field records the total byte length of the entire multidimensional pressure matrix data structure, used by the receiver to determine if the data has been received completely. The check field uses a cyclic redundancy check (CRC) code, whose calculation range covers all bytes of the timestamp, row number identifier, column number identifier, and pressure value sequence, used by the receiver to verify whether any errors occurred during data transmission. The timestamp is the absolute time when the microcontroller started the current frame row scanning operation, using microsecond-level timing precision, used to identify the acquisition time of the current frame pressure data, providing a time reference for the timing analysis of subsequent continuous frame data. The row number identifier and column number identifier record the total number of rows and columns of the sensor nodes corresponding to the current frame pressure matrix, respectively, used by the receiver to determine the row and column address identifiers corresponding to each pressure value when parsing the pressure value sequence. The pressure value sequence is arranged in row-major order, that is, the pressure values ​​corresponding to all column sensing nodes in the first row are arranged first, then the pressure values ​​corresponding to all column sensing nodes in the second row are arranged, and so on. Each element in the pressure value sequence corresponds to a unique row and column address identifier of the flexible pressure sensing node. The receiving end can calculate the row and column address identifier corresponding to the element through the row number identifier, column number identifier and the position of the element in the sequence, thus completing the matching of pressure value and spatial location.

[0027] For the pressure value sequence in the multidimensional pressure matrix data structure, spatial filtering and spline interpolation reconstruction are performed. Spatial filtering constructs a sliding window of a preset size centered on the target element in the current pressure value sequence. The spatial distance weights of each element within the sliding window relative to the central element are calculated. These spatial distance weights are then used to perform a weighted summation of the elements within the sliding window, generating filtered element values ​​that replace the target element values. The size of the sliding window can be configured according to the distribution density of the sensor nodes. The spatial distance weights within the window are negatively correlated with the Euclidean distance of the element relative to the central element; elements closer to the central element have larger spatial distance weights, and elements farther away have smaller spatial distance weights. This weighted summation method suppresses random noise in the pressure value sequence while preserving the spatial characteristics of the pressure distribution. The calculation process for spatial filtering is implemented using the following formula:

[0028] in, For the filtered first Line number The pressure value of the column; The width of the sliding window is half its width, and the total size of the sliding window is [size missing]. ; This represents the original pressure value at the corresponding position within the sliding window; The spatial distance weights are the corresponding positions within the sliding window. , To avoid extremely small constants with a denominator of 0.

[0029] For missing positions in the filtered pressure value sequence caused by sensor node damage, line faults, etc., cubic spline interpolation is performed using multiple valid elements adjacent to the missing position as interpolation nodes to fill the missing positions. The cubic spline interpolation calculation uses the valid pressure values ​​of the row or column containing the missing position as interpolation nodes to construct a piecewise cubic polynomial function. This ensures the continuity of the function values, first derivative, and second derivative at the interpolation nodes, achieving smooth reconstruction of the pressure value sequence, supplementing missing pressure data, and ensuring the integrity of pressure distribution data in subsequent analysis. The cubic spline interpolation process is implemented using the following formula:

[0030] in, It is a cubic spline interpolation function; , The x-coordinates of two adjacent interpolation nodes; , , , For the first The spline coefficients within the interpolation interval satisfy the boundary conditions of continuous function values, continuous first derivatives, and continuous second derivatives at the interpolation nodes.

[0031] In this embodiment, all fields of the multidimensional pressure matrix data structure are standardized and defined using the table below to ensure that the data sender and receiver have unified parsing rules.

[0032] Table 2. Field Definitions of the Multidimensional Pressure Matrix Data Structure

[0033] The table above defines a standardized encapsulation format for dynamic pressure distribution data. The microcontroller organizes the collected pressure data according to the field order and byte length defined in the table. After receiving the data, the edge computing module parses and verifies the data frame according to the definition in the table to avoid parsing errors during data transmission and ensure the consistency of data interaction.

[0034] This embodiment refines the node structure and data encapsulation format of the pressure sensing network. It achieves spatial positioning of pressure data through unique row and column address identifiers, ensures data transmission consistency by utilizing a standardized multidimensional pressure matrix data structure, and improves the signal-to-noise ratio and integrity of pressure data through spatial filtering and cubic spline interpolation, providing high-quality input data for subsequent attitude recognition and analysis.

[0035] In another preferred embodiment, reference Figure 3 The microcontroller incorporates a polling scheduler and a dynamic frequency register. The polling scheduler divides the high-resolution matrix flexible pressure sensor network into multiple scanning sub-regions, while the dynamic frequency register stores the polling cycle value corresponding to each scanning sub-region. The polling scheduler is stored in the microcontroller's non-volatile memory. Each scanning sub-region corresponds to a continuous series of rows and columns of sensor nodes. All scanning sub-regions are non-overlapping and cover the entire effective sensing area of ​​the sensor network. Each scanning sub-region corresponds to a unique scheduling entry in the polling scheduler, which includes the row address range, column address range, initial polling cycle value, and scheduling priority for that sub-region. The dynamic frequency register is a read-write register group within the microcontroller. Each register in the register group corresponds to a scanning sub-region, storing the currently active polling cycle value for that sub-region. The value in the register is dynamically updated based on pressure changes within the scanning sub-region.

[0036] The microcontroller's time-polling mechanism triggers read operations on each scan sub-region sequentially according to the scheduling priority in the polling schedule table. Within the read interval of a single scan sub-region, it executes data parsing tasks for other scan sub-regions. Specifically, based on the scheduling order in the polling schedule table, the microcontroller first completes read operations on all sensor nodes in the highest-priority scan sub-region. Then, within the interval before the next read operation in that scan sub-region is triggered, it executes data parsing tasks for other scan sub-regions that have already been read. These tasks include analog-to-digital conversion, data verification, row and column address matching, and data frame encapsulation, enabling parallel processing of data reading and parsing, thus improving the microcontroller's operating efficiency and the real-time performance of data acquisition.

[0037] When the pressure change rate within any scanned sub-region exceeds a preset threshold, the microcontroller updates the polling period value in the dynamic frequency register, shortening the polling period for the corresponding scanned sub-region. The pressure change rate of the scanned sub-region is the relative change between the average pressure value of that region in the current frame and the average pressure value of that region in the previous frame. The pressure change rate is calculated using the following formula:

[0038] in, For the first The scanning sub-region in the first The rate of change of pressure corresponding to the frame; For the first The scanning sub-region in the first The average pressure value of the frame; For the first The scanning sub-region in the first The average pressure value of the frame; To avoid extremely small constants with a denominator of 0.

[0039] During the process of updating the polling cycle value in the dynamic frequency register of the microcontroller, historical time-series pressure change rate is incorporated into the calculation. The microcontroller maintains a corresponding circular queue of change rates for each scan sub-region. The length of the circular queue is fixed, and it stores the pressure change rate values ​​for multiple consecutive frames of the corresponding scan sub-region. The microcontroller pushes the pressure change rate obtained in the current polling cycle to the tail of the corresponding circular queue, while simultaneously removing historical data from the head of the queue, ensuring that the queue always stores the latest fixed number of pressure change rate values. A moving average is calculated for all pressure change rates in the circular queue to generate a trend change rate. The trend change rate is calculated using the following formula:

[0040] in, For the first The scanning sub-region in the first The rate of change of the trend corresponding to the frame; The fixed length of the rate-of-change circular queue; For the first The scanning sub-region in the first The rate of change of pressure corresponding to the frame.

[0041] The microcontroller calculates the absolute value of the difference between the current pressure change rate and the trend change rate. Based on this absolute value, it queries a pre-stored frequency adjustment step size mapping table within the microcontroller to obtain the corresponding frequency adjustment step size. This step size is then used to modify the polling cycle value in the dynamic frequency register. A larger absolute value corresponds to a larger frequency adjustment step size, resulting in a greater reduction in the polling cycle and a higher data acquisition frequency for the corresponding scanned sub-region. Conversely, a smaller absolute value corresponds to a smaller frequency adjustment step size, leading to a smaller adjustment range in the polling cycle and preventing system instability caused by frequent fluctuations in the polling cycle. The microcontroller stores the minimum and maximum polling cycle values ​​for each scanned sub-region. The adjusted polling cycle value always falls between these minimum and maximum values, preventing excessive resource consumption due to excessively shortened polling cycles or data acquisition lag due to excessively extended cycles.

[0042] In this embodiment, the polling cycle operating parameters for each scanned sub-region are configured using the table below, providing boundary constraints for the dynamic adjustment of the polling cycle.

[0043] Table 3. Configuration Table of Scan Sub-region Polling Cycle Parameters

[0044] The table above configures the operating parameter boundaries of the polling cycle for each scanning sub-region. All parameters can be adjusted according to the application scenario of the sliding pad and the patient's physical condition to avoid system instability caused by excessive adjustment of the polling cycle. At the same time, different initial parameters are configured for scanning sub-regions at different locations to adapt to the pressure change characteristics of different areas of the sliding pad.

[0045] This embodiment refines the time-polling acquisition mechanism of the microcontroller. By dividing the scanning sub-regions and configuring the polling schedule table, parallel processing of data reading and parsing is achieved. At the same time, combined with the trend analysis of historical time-series pressure change rate, the polling cycle of the corresponding scanning sub-region is dynamically adjusted. In areas of sudden pressure changes, the data acquisition density is increased, and in areas of stable pressure, the system resource consumption is reduced, thereby achieving a balance between acquisition efficiency and system load.

[0046] In yet another preferred embodiment, reference is made to Figure 4 The edge computing module's built-in posture recognition algorithm models the data on pressure area and duration. This process includes binarizing the dynamic pressure distribution data to generate a pressure mask matrix, and then performing an eight-connected component labeling algorithm on the pressure mask matrix to extract at least one connected pressure-bearing region. During binarization, matrix elements with pressure values ​​exceeding a preset pressure threshold are set to 1, while those with pressure values ​​below the threshold are set to 0. This generates a binarized pressure mask matrix with the same dimensions as the dynamic pressure distribution data. Elements with a value of 1 in the pressure mask matrix correspond to the pressure-bearing positions where the patient is in contact with the sliding pad, while elements with a value of 0 correspond to non-contact positions. The eight-connected component labeling algorithm traverses all elements in the pressure mask matrix. For elements with a value of 1, it checks their adjacent elements in the top, bottom, left, right, and four diagonal directions, marking adjacent elements with a value of 1 as the same connected component. After traversal, all marked connected components are extracted as connected pressure-bearing regions, each corresponding to an independent pressure point where the patient's body is in contact with the sliding pad.

[0047] For each connected compressed region, the number of pixels within the region is counted, and the compressed area is obtained by multiplying the number of pixels by the area of ​​a single pixel. The calculation of the compressed area is achieved using the following formula:

[0048] in, For the first The pressure area of ​​each connected pressure zone; For the first The number of valid pixels in each connected compressed region; This represents the actual physical contact area corresponding to a single pressure sensing node, and is a fixed constant.

[0049] A duration accumulator is established corresponding to each connected pressure-bearing region. The initial count value of the duration accumulator is 0. The overlap rate of the connected pressure-bearing regions is compared in the dynamic pressure distribution data of consecutive frames. When the overlap rate is greater than a set proportion, the duration accumulator is triggered to accumulate the count and generate the pressure duration. The calculation process of the overlap rate of the connected pressure-bearing regions between consecutive frames is implemented by the following formula:

[0050] in, For the first The first frame The connected pressure zone and the first The overlap rate of corresponding connected compressed regions in the frame; For the first The first frame The area of ​​each connected pressure zone; For the first The area of ​​the corresponding connected compressed region in the frame; This is for area intersection operations; This is an area union operation. When the overlap rate is greater than a set ratio, it is determined that the connected compressed region of the current frame and the connected compressed region of the previous frame are the continuous compressed regions corresponding to the same patient's body part. The corresponding duration accumulator is triggered to accumulate the count. The accumulated count value is the time interval between the current frame and the previous frame. The accumulated count value of the duration accumulator is the continuous compression duration of the connected compressed region.

[0051] During the process of counting and accumulating to generate the pressure duration in the duration accumulator, a weight attenuation correction operation based on the geometric deformation parameters of the connected pressure region is added. For each connected pressure region, the aspect ratio of the bounding rectangle of the corresponding region in the pressure mask matrix and the region area dispersion index are calculated. The aspect ratio of the bounding rectangle is the ratio of the width to the height of the smallest bounding rectangle of the connected pressure region, and the region area dispersion index is the ratio of the area of ​​the smallest bounding rectangle of the connected pressure region to the actual area of ​​the connected pressure region. A weight attenuation function is constructed with the aspect ratio and the region area dispersion index as independent variables. The calculation result of the weight attenuation function is used as the attenuation coefficient and multiplied into the single count value of the duration accumulator. The calculation process of the weight attenuation function is implemented through the following formula:

[0052] in, For the first The first frame The attenuation coefficient corresponding to each connected compressed region; For the first The first frame The aspect ratio of the circumscribed rectangle of each connected pressure region; The default standard aspect ratio; For the first The first frame The area dispersion index of a connected compressed region; , These are preset weighting coefficients, all of which are non-negative constants. When the aspect ratio deviates from the preset standard range or the area dispersion index increases, the attenuation coefficient decreases, and the cumulative amplitude of a single count decreases, thereby correcting the duration of pressure application. This eliminates the interference of geometric deformation of the pressure area caused by the patient's turning over or limb movement on the statistics of pressure application duration, and avoids false triggering of high pressure ulcer risk alarms.

[0053] refer to Figure 5 The attitude recognition algorithm's process of performing in-depth analysis of slip displacement trends includes extracting multi-frame dynamic pressure distribution data within a continuous time window and calculating the global pressure centroid coordinates for each frame of dynamic pressure distribution data. The calculation of the global pressure centroid coordinates is achieved using the following formula:

[0054] in, For the first The global pressure centroid coordinates corresponding to the frame dynamic pressure distribution data; This represents the total number of rows in the pressure sensor network. This represents the total number of columns in the pressure sensor network. For the first The first frame Line number Pressure values ​​of the sensor nodes; For the first Line number The physical coordinates of the column sensor nodes.

[0055] The global pressure centroid coordinates of two adjacent frames are subtracted to generate the centroid displacement vector. The calculation of the centroid displacement vector is achieved through the following formula:

[0056] in, For the first Frame relative to the first The centroid displacement vector of the frame; , For the first The horizontal and vertical coordinates of the centroid of the frame; , For the first The centroid's horizontal and vertical coordinates within a frame. Multiple centroid displacement vectors within a continuous time window are arranged in chronological order to construct a centroid trajectory matrix. A first-order difference operation is performed on the centroid trajectory matrix to generate the velocity gradient matrix. The calculation of the velocity gradient matrix is ​​achieved using the following formula:

[0057] in, The velocity gradient matrix; The starting frame number of the continuous time window; This is the sequence number of the last frame of the continuous time window; For the first The centroid displacement vector of the frame. Extract continuous vector segments from the velocity gradient matrix that exceed a preset velocity threshold, and output them as the sliding displacement trend. The time length and magnitude of the continuous vector segments correspond to the duration and amplitude of the patient's sliding motion.

[0058] The alarm data packets generated by the edge computing module are assembled according to a preset encapsulation protocol. This protocol defines a start symbol, device identifier, alarm type field, characteristic payload field, and end symbol. The start and end symbols are fixed values ​​used to identify the start and end positions of the alarm data packet. The device identifier is a unique device number for the sliding pad control system, used by the nursing station to identify the source device of the alarm data. The alarm type field is filled with a high-pressure ulcer risk indicator or an abnormal sliding indicator to identify the type of alarm event. The characteristic payload field is filled with the corresponding pressure area value, pressure duration value, or sliding displacement trend vector when the alarm is triggered, used by the nursing station to view detailed characteristic data of the alarm event. After receiving the alarm data packet, the wireless gateway matches it to a preset transmission channel queue according to the alarm type field, and segments the alarm data packet into multiple fragmented data packets with sequence numbers for transmission. Different alarm types correspond to different priority transmission channel queues. High priority queues have higher transmission bandwidth and lower transmission latency. The length of the fragmented data packet does not exceed the maximum transmission unit of the corresponding transmission channel. Each fragmented data packet carries a continuous sequence number. The receiving end reassembles the fragmented data packets into a complete alarm data packet according to the sequence number. At the same time, it performs retransmission operations for lost fragmented data packets to ensure the integrity of alarm data transmission.

[0059] refer to Figure 6 The slip displacement trend and the characteristics of the pressure-bearing region are fused using tensors and mapped into control commands for a micro-airbag solenoid valve array. The mean direction and magnitude of continuous vector segments in the velocity gradient matrix are extracted, and combined with the geometric center coordinates of the pressure-bearing region extracted from the pressure mask matrix, a three-dimensional feature tensor containing directional, magnitude, and positional features is constructed. The construction process of the three-dimensional feature tensor is achieved through the following formula:

[0060] in, It is a three-dimensional feature tensor; This is the unit vector representing the mean direction of the slip displacement trend; This represents the average modulus value corresponding to the slip displacement trend; The coordinates of the geometric center of the pressure zone are given. The three-dimensional feature tensor is input into a pre-defined fully connected mapping network, which outputs a sequence of switching states and inflation / deflation durations corresponding to each solenoid valve node in the micro-airbag solenoid valve array. These sequences serve as feedback control signals. The calculation process of the fully connected mapping network is achieved through the following formula:

[0061] in, This is the solenoid valve switching state sequence; This is a sequence of the charging and discharging durations of the solenoid valve. , These are the weight matrices for the hidden layer and the output layer, respectively. , These are the bias vectors for the hidden layer and the output layer, respectively; The activation function for the hidden layer; The piecewise activation function for the output layer outputs discrete values ​​of 0 or 1 for the switch state sequence, corresponding to the closed or open state of the solenoid valve. For the inflation / deflation duration sequence, it outputs a non-negative continuous value, corresponding to the duration the solenoid valve is open. The weight matrix and bias vector of the fully connected mapping network converge through pre-training. The pre-training process uses pressure distribution data from different patients and corresponding airbag adjustment strategies as training datasets to ensure that the control commands output by the network can achieve a balanced pressure distribution on the patient's contact surface and limit abnormal slippage.

[0062] In this embodiment, the mapping relationship between alarm types and transmission channel queues is defined by the following table, providing standardized rules for the transmission of alarm data packets.

[0063] Table 4. Alarm Type Field Mapping Table with Transmission Channel Queue

[0064] The table above defines the transmission rules corresponding to different alarm types. High-priority transmission channels are allocated to high-risk alarm events to ensure reliable and timely transmission of alarm data and avoid loss or delay of alarm information. At the same time, differentiated transmission parameters are allocated to different types of data to achieve reasonable allocation of transmission resources.

[0065] This embodiment refines the complete implementation logic of the attitude recognition algorithm. It achieves accurate statistics of the pressure area and pressure duration by marking eight connected domains and accumulating the duration. It eliminates the interference of geometric deformation by combining weight decay correction. It identifies the slip displacement trend by analyzing the centroid trajectory and velocity gradient. At the same time, it standardizes the encapsulation and transmission mechanism of alarm data packets. It achieves precise automatic control of the airbag solenoid valve array by fusing and mapping three-dimensional feature tensors, thereby improving the system's accuracy in identifying abnormal states and the accuracy of control response.

Claims

1. A control system for a medical multifunctional mobile sliding mat, characterized in that, This includes a high-resolution matrix flexible pressure sensor network, microcontroller, edge computing module, wireless gateway, and miniature airbag solenoid valve array laid inside the sliding pad; The microcontroller continuously collects dynamic pressure distribution data of the patient contact surface output by the high-resolution matrix flexible pressure sensor network using a preset time polling mechanism. The edge computing module has a built-in posture recognition algorithm, receives the dynamic pressure distribution data, and performs data modeling and in-depth analysis on the patient's pressure area, pressure duration, and slippage trend. The process of deep analysis of the slip displacement trend by the attitude recognition algorithm includes: extracting multi-frame dynamic pressure distribution data within a continuous time window, and calculating the global pressure centroid coordinates of each frame of dynamic pressure distribution data. The global pressure centroid coordinates of two adjacent frames are subtracted to generate a centroid displacement vector. Multiple consecutive centroid displacement vectors are arranged in chronological order to construct a centroid trajectory matrix. The centroid trajectory matrix is ​​subjected to a first-order difference operation to generate a velocity gradient matrix, and continuous vector segments exceeding a preset velocity threshold in the velocity gradient matrix are extracted as slip displacement trend output. When a risk of pressure sores is detected or an unexpected abnormal slip occurs, the edge computing module sends an alarm data packet to the nursing station through the wireless gateway, and at the same time outputs a feedback control signal based on the force model. The control system of the medical multifunctional mobile sliding pad, which receives the feedback control signal from the micro airbag solenoid valve array, performs zoned inflation or deflation operations.

2. The control system for the medical multifunctional mobile sliding mat according to claim 1, characterized in that, The high-resolution matrix flexible pressure sensor network consists of flexible pressure sensing nodes with alternating rows and columns, and each flexible pressure sensing node is assigned a unique row and column address identifier. The dynamic pressure distribution data is organized into a multi-dimensional pressure matrix data structure, which includes a frame header, timestamp, row number identifier, column number identifier, and a pressure value sequence arranged in row priority order. The frame header includes a data length field and a check field. The timestamp is the absolute time when the microcontroller reads the flexible pressure sensor node. Each element in the pressure value sequence corresponds to the unique row and column address identifier.

3. The control system for the medical multifunctional mobile sliding mat according to claim 1, characterized in that, The microcontroller includes a polling schedule table and a dynamic frequency register. The polling schedule table divides the high-resolution matrix flexible pressure sensor network into multiple scanning sub-regions, and the dynamic frequency register stores the polling period value corresponding to the scanning sub-region. The time polling mechanism triggers reading operations on each scanned sub-region sequentially according to the polling schedule table, and performs data parsing tasks for other scanned sub-regions within the reading interval of a single scanned sub-region. When the rate of pressure change in any scanned sub-region exceeds a preset change threshold, the microcontroller updates the polling cycle value in the dynamic frequency register to shorten the polling cycle of the corresponding scanned sub-region.

4. The control system for the medical multifunctional mobile sliding pad according to claim 1, characterized in that, The posture recognition algorithm's data modeling process for the pressure area and pressure duration includes: binarizing the dynamic pressure distribution data to generate a pressure mask matrix, and performing an eight-connected domain labeling algorithm on the pressure mask matrix to extract at least one connected pressure region; For each connected compressed region, the compressed area is obtained by multiplying the number of internal pixels by the area of ​​a single pixel. Establish a duration accumulator corresponding to each of the connected pressure areas. Compare the overlap rate of the connected pressure areas in the dynamic pressure distribution data of multiple consecutive frames. When the overlap rate is greater than a set ratio, trigger the duration accumulator to count and accumulate to generate the pressure duration.

5. The control system for the medical multifunctional mobile sliding pad according to claim 1, characterized in that, The alarm data packets generated by the edge computing module are assembled according to a preset encapsulation protocol, which defines a start symbol, a device identifier, an alarm type field, a feature load field, and an end symbol. The alarm type field is filled with a high pressure sore risk indicator or an abnormal sliding indicator, and the characteristic load field is filled with the corresponding value of the pressure area, the pressure duration, or the sliding displacement trend vector when the alarm is triggered. After receiving the alarm data packet, the wireless gateway matches the alarm type field with a preset transmission channel queue, and divides the alarm data packet into multiple fragmented data packets with sequence numbers for transmission.

6. The control system for the medical multifunctional mobile sliding pad according to claim 2, characterized in that, Spatial filtering and spline interpolation reconstruction are performed on the pressure value sequence in the multidimensional pressure matrix data structure. The spatial filtering process constructs a sliding window of a preset size centered on the target element in the current pressure value sequence, calculates the spatial distance weight of each element within the sliding window, and uses the spatial distance weight to perform a weighted summation of the elements within the sliding window to generate filtered element values, which then replace the target element values. For missing positions in the filtered pressure value sequence caused by sensor node damage, cubic spline interpolation is performed using multiple valid element values ​​adjacent to the missing position as interpolation nodes, and the interpolation results are filled into the missing positions.

7. The control system for the medical multifunctional mobile sliding pad according to claim 3, characterized in that, The historical time-series pressure change rate is incorporated into the calculation during the process of updating the polling cycle value in the dynamic frequency register by the microcontroller. The microcontroller maintains a circular queue of change rates corresponding to each scanning sub-region, pushes the pressure change rate obtained in the current polling cycle into the tail of the corresponding circular queue of change rates and removes the data at the head of the queue. A trend rate of change is generated by performing a moving average calculation on all pressure change rates in the rate of change cycle queue. The absolute value of the difference between the current pressure change rate and the trend rate of change is used to look up the frequency adjustment step size in the table. The polling period value in the dynamic frequency register is then modified by superimposing the frequency adjustment step size.

8. The control system for the medical multifunctional mobile sliding pad according to claim 4, characterized in that, In the process of counting and accumulating to generate the pressure duration in the duration accumulator, a weight attenuation correction operation based on the geometric deformation parameters of the connected pressure region is added; For each connected pressure region, calculate the aspect ratio of the bounding rectangle of the corresponding region in the pressure mask matrix and the region area dispersion index; Construct a weighted decay function with aspect ratio and regional area dispersion index as independent variables, and multiply the calculation result of the weighted decay function as the decay coefficient into the single count value of the duration accumulator. The aspect ratio deviates from the preset standard range or the area dispersion index increases.

9. The control system for the medical multifunctional mobile sliding mat according to claim 1, characterized in that, The slip displacement trend and the characteristics of the pressure area are fused using tensors and mapped into control commands for a micro airbag solenoid valve array. The mean direction and magnitude of the continuous vector segments in the velocity gradient matrix are extracted, and combined with the geometric center coordinates of the pressure region extracted from the pressure mask matrix, a three-dimensional feature tensor containing directional features, magnitude features and position features is constructed. The three-dimensional feature tensor is input into a preset fully connected mapping network, and the output is a sequence of switching states and a sequence of inflation and deflation durations corresponding to each solenoid valve node in the micro airbag solenoid valve array, which serve as feedback control signals.

Images